I just went to UberConf 2016. As always Jay Zimmerman puts on a great show. As great as UberConf is, it’s probably a conference you can get to every other year. Lots of familiar speakers, they’re all great. I was surprised to see so few people I know in attendance. Maybe people come to their senses and get into management and I’m still slinging code.
I took the all day workshop on reactive programming given by Venkat. It gave me a good basis for reactive programming. Venkat gave an entertaining presentation as always and he covered a lot of ground.
My hangup would be that he focused on the idea of streams or data-flow computing. He gave examples like stock data or telemetry data continually streaming into your application to be processed. I’ve been coding professionally for almost 20 years and I have never been in a domain where that happens. Honestly that kind of turned me off. It’s not sexy but the fact is developers primarily create HTTP front-ends for either a persistence or some other back-end system. From Venkat’s perspective if your app doesn’t deal with streams of information then it’s not going to be of much use.
Spring and Cloud
I recently have become re-interested in Spring. I used Scala for a few years and then Groovy/ Grails. I forgot how useful all the Spring frameworks were. They have kept up with things and a lot of the battle of being productive is just being aware of what’s out there.
I did a lot of by Craig Walls, they were great. The theme of Craig’s talks were centered around cloud native. I like the idea, the point is that there are some new tools and concerns we’ll need when going to the cloud. It doesn’t mean that all this is exclusive just to the cloud. There was mention of 12 Factor Application I’ve gone through it before, seems like it’s something I need to re-visit.
He covered a lot of things about Spring. Spring is so huge and frankly they do a so-so job of organizing, documenting, deprecating and communicating best practices. There’s just so much it’s probably very hard for them to keep a handle on it. My point is that it’s great to hear someone speak that is up to speed on the best practices, how everything is positioned and really just letting you know what’s out there.
I also went to a lot of Matt Stine talks about microservices. In covering microservices he also touched on Spring frameworks that help you. Matt also covered reactive programming in a more pragmatic way such that I could imagine using it with real work.
That presentation went on about how great Netflix is, good on them, they’re doing amazing stuff. But one thing that rubbed me wrong is how proud they are that they only hire senior engineers. Is that a good thing if everyone just hires senior engineers?
I did a 2 session Docker workshop. There were a lot of people in attendance and any time you’re doing virtualization with more than 100 people things go off the rails quick. So Chrisopher Judd had his work cut out for him. I was glad I attended, for whatever reason I couldn’t fit some Docker concepts into my head just from reading but the sessions were useful.
- There’s an interesting relationship with the cloud and microservices. If you want to have microservices you probably want to do cloud things to manage them even if you don’t use a cloud.
- Spring and Netflix are doing great things, if you’re doing microservices or cloud stuff on the JVM you probably want to get familiar with the opensource tools they have available.
- If you want to leverage Twitter’s Zipkin you can use Spring Cloud Sleuth.
- At least right now it’s too soon for me to poke around with Java9 module capabilities. It’ll be interesting when there is build system and tooling support.
- Use Netflix Eureka for service discovery. This prevents you from hard-coding URLs to services your application depends on.
- Use Netflix Zuul to dynamically route services.
- Zuul could be used to host static assets and be a gateway for services. If you do this then you don’t have to worry about cross origin request concerns.
- Netflix has an interesting approach to load balancing, Ribbon runs on the client, is made aware of services by Eureka and does a sort of load balancing in the client. There isn’t a load balancer that sits in front of services. So, the client using Ribbon will round-robin or lay off of poorly behaving services for you.
- Use Spring Cloud Config to manage configuration in your cloud applications.
- ELK seems to be the approach for distributed, searchable logging. Seems like a bunch of work to get going, we use Splunk at work.